Les propriétés des sols et leur mesure - ISSMGE [PDF]

Section 1

Les propriétés des sols et leur mesure Soil Properties and their Measurement

Sujets de discussion : Dispersion des essais de mécanique des sols. Dissipation de la pression interstitielle. Propriétés visco-élastiques des sols : Existence d ’un seuil de Bingham. Fluage. Subjects for Discussion: Scatter of soil mechanics tests. Pore-pressure dissipation. Visco-elastic properties of soils : the existence of a Bingham limit. Creep.

Président / Chairman : R. G lo sso p , Grande Bretagne Vice-Président / Vice-Chairman : P. W a h l, France Rapporteur Général / General Reporter : C. M e y e rh o f, Canada Membres du Groupe de Discussion / Members of the Panel: A.W. B ishop, Grande Bretagne; H . B r e th , Allemagne; N. D enissov, U.R.S.S.; E. G eu ze, Etats-Unis ; R. M a rc h a n d , France. Discussion orale / Oral discussion P. Anagnosti, Yougoslavie. S. Andrei, Roumanie. R. A. Ashbee, Grande-Bretagne. A. W. Bishop, Grande-Bretagne. H. Breth, Allemagne. J. Coleman, Grande-Bretagne. N. Denissov, U.R.S.S. E. C. W. Geuze, U.S.A. A. Kezdi, Hongrie. T. W. Lambe, U.S.A. G. A. Leonards, U.S.A. R. Marchand, France. L. Ménard, France. J. Osterman, Suède. B. V. Ranganatham, Inde. K. Roscœ, Grande-Bretagne. P. W. Rowe, Grande-Bretagne. L. Suklje, Yougoslavie. G. Wiseman, Israël.

Contributions écrites / Written Contributions S. Andrei, Roumanie. R. A. Ashbee, Grande-Bretagne. J. Bernède, France. H. Borowicka, Autriche. J. B. Burland, Afrique du Sud. T. K. Chaplin, Grande-Bretagne. D. H. Cornforth, Grande-Bretagne. C. B. Crawford, Canada. R. Davidenkoff, Allemagne. R. Di Martino, Italie. Ventura Escario, Espagne. V. A. Florin et V. P. Sipidin, U.R.S.S. B. P. Gorbunov, U.R.S.S. I. M. Gorkova, N. A. Oknina et V. F. Chepik, U.R.S.S. W. Heierli, Suisse. W. M. Kirkpatrick, Grande-Bretagne. J. L. Kogan U.R.S.S. B. Ladanyi, Yougoslavie. T. E. Phalen Jr., U.S.A. A. Piaskowski, Pologne. P. W. Rowe, Grande-Bretagne. H. U. Smoltczyk, Allemagne. E. Spencer, Grande-Bretagne. G. Stefanoff, Bulgarie. Tan Tjong-Kie, Chine. F. Terracina, Italie. K. Terzaghi, U.S.A. W. J. Thompson, Grande-Bretagne. A. H. Toms, Grande-Bretagne. C. Veder, Italie. S. S. Vialov, U.R.S.S. P. Wikramaratna, Ceylan. G. S. Zolotarev, U.R.S.S. Le Vice-Président :

Messieurs, au nom du Comité d’organisation- du Congrès, j ’ai l’honneur d’ouvrir la première séance de discussion du Cinquième Congrès de la Société internationale de Méca­ nique des sols, séance consacrée à la section 1. « Propriétés des sols et leur mesure ». Je prie M. Glossop, de la délégation britannique de bien C. M ey er h o f Rapporteur General, Division 1 / General Reporter, Division 1 vouloir assurer la présidence de cette section, assisté de 87

M. Meyerhof, de la délégation du Canada, comme rappor­ teur, de M. Bishop, de la délégation britannique, de M. Breth, de la délégation allemande, de M. Denissov, de la délégation soviétique, et de M. Marchand de la délégation française, comme membres du groupe de discussion. M. le président va vous indiquer la façon dont la séance va se dérouler. Le Président : This morning we are to discuss soil properties and their measurement. This is, of course, a very large and complex subject, and so as to give continuity to our proceedings the Organising Committee has suggested that the spoken discus­ sion should be limited to three aspects of the subject. These, as you know from page 37 of the Programme are — (a) Scatter of soil mechanic tests; (b) Pore-pressure dissipation; (c) Visco­ elastic properties of soil; the existence of a Bingham limit. Creep. Accordingly I propose that we should give priority to these three subjects, though if time permits other subjects may be introduced later. Of course, written contributions on any subject connected with Section 1 may be submitted and will be published in Volume III. These should not exceed 600 words in length and they should be submitted within one month of the closing of this Conference. As regards the order of our discussion, I will first call upon Mr Meyerhof to make his report. The discussion will then be opened by the panel, or discussion group, whose names have been given to you by Mr Wahl. They will discuss these three subjects in turn (that is to say, they will first discuss (a), then (b) and then (c)). They will be allowed 10 to 15 min­ utes each. After the panel has introduced the subject in this way we will have a short break of 10 minutes or so, and during that 10 minutes if any other members — and I hope there will be many — of the audience wish to contribute will they please write their names and the subject they wish to discuss (that is, (a), (b) or (c)) on a piece of paper and send it up to the platform. Mr Meyerhof and I will, during the interval, go through these and call upon the members to speak. In making their discussion they should come up to the tribune and speak through the microphone here. I would also like to remind them to speak slowly and clearly, since every word they say must, of course, be interpreted and tape recorded. I am afraid that, again, time must be limited because no doubt many people will wish to take part. Therefore I ask members to speak for not more than 5 minutes. I now ask Mr Meyerhof to give his report. Le Rapporteur Général : Monsieur le Président, Mesdames, Messieurs, j ’ai présenté en anglais le rapport général de cette section, et je voudrais vous en faire un résumé en français, pour marquer le carac­ tère international de cette conférence, et pour remercier nos hôtes de leur aimable hospitalité. La première section du Congrès de Mécanique des Sols est consacrée aux propriétés des sols et à leur détermination. Il s’agit d’un des problèmes fondamentaux de notre science, et dont l’importance est soulignée par l’envoi de soixantedouze communications, le plus grand nombre jamais obtenu dans aucune autre section. Mon rapport a donc été basé principalement sur l’analyse des communications au présent Congrès; les publications parues depuis le dernier Congrès de Londres sont brièvement mentionnées. On peut distinguer trois grands problèmes concernant les propriétés des sols : classification et description; propriétés physico-chimiques et mécaniques, y compris la perméabilité; méthodes et appareils de mesure des propriétés des sols. Ces questions ont été étudiées dans les sept sections princi­ pales de notre rapport de manière à faire ressortir le compor­

tement fondamental des sols, qui a donné lieu, récemment, à une recherche intense sut l’aspect physico-chimique, du cisaillement et de la déformation. Néanmoins, je suggère que pour nos conférences prochaines cette section soit divisée en une section 1A concernant la classification et description des sols avec les méthodes et appareils correspondants, et une section 1B concernant les propriétés physico-chimiques et mécaniques, y compris la perméabilité, avec les méthodes et appareils correspondants. Voici un bref résumé de mon rapport général : Dans plusieurs pays les études régionales des sols peuvent être groupées, et les renseignements obtenus utilisables sous forme de cartes géotechniques et de mémoires pour les avantprojets d’ouvrages en terre et de travaux de fondations dans la région même. On n ’observe pas de contribution majeure dans l’identification et la classification des sols, mais des rapprochements nouveaux ont été proposés entre les propriétés descriptives et mécaniques. La nature et le comportement du complexe eau-sol ont été précisés par des méthodes physico­ chimiques dans les domaines des échanges de bases, de la sta­ bilisation, et du gonflement. On note également un progrèdans l’étude de Félectro-osmose, du gel, et de la perméas bilité d’un sol partiellement saturé; néanmoins les principes de base des effets constatés et les modalités de l’application pratique ne sont pas encore établis. Les propriétés visco-élastiques des sols sont mieux connues, et les caractéristiques de déformation déterminées à con­ trainte constante et dans le cas de répétition des efforts, mais fréquemment sans drainage et sans mesure des pressions interstitielles; et les propriétés des sols drainés, primordiales pour les problèmes de stabilité à long terme, ne sont pas connues suffisamment. La plupart des essais présentés ren­ voient à des contraintes triaxiales, et les relations viscoélastiques dans un champ de contraintes biaxiales sont mécon­ nues. La consolidation axiale et radiale a été étudiée pour des sols partiellement ou complètement saturés et également dans le cas d’argiles anisotropes; la compressibilité des sols pulvérulents a été envisagée, le tassement étant jusqu’à main­ tenant basé sur des méthodes semi-empiriques. C’est dans le domaine de la résistance au cisaillement qu’on trouve le plus grand nombre de communications, relatives à des considérations théoriques et expérimentales diverses. Les paramètres les plus utiles concernent la résis­ tance effective, et nombreux sont les mémoires basés sur des essais triaxiaux. La résistance au cisaillement avec drainage et à long terme, de même que les paramètres résultant de compressions biaxiales ou d ’autres modes de chargement, devraient être étudiés davantage. L’appareillage de compres­ sion triaxiale et de consolidation a été amélioré, et de nou­ velles méthodes de mesure de la pression interstitielle ont été mises au point. Beaucoup d’intéressantes communications invitent à la discussion, mais les sujets doivent être limités à trois pro­ blèmes importants : 1. La dispersion des essais de Mécanique des Sols, le degré de dispersion devant être étudié d’abord sur une propriété donnée d’un certain sol. Il importe de distinguer d’une part la dispersion due à la diversité des appareils, des opé­ rateurs, et des techniques d’essais et d ’interprétation, et d’autre part celle due aux variations du sol lui-même. 2. La dissipation de la pression interstitielle, qui peut être estimée dans le cas de matériaux isotropes complètement saturés et consolidés sous une charge constante ou unifor­ mément variée. Il serait désirable d’envisager les cas d’autres types de chargement, de sols anisotropes, et de sols partiellement saturés où interviennent à la fois les pressions de l’air et de l’eau. 3. Les propriétés visco-élastiques des sols : existence d’un seuil de Bingham; problème du fluage. Vu l’importance

des propriétés de déformation visco-élastique et de résis­ tance à long terme, il serait utile de considérer la réparti­ tion du seuil de Bingham entre la cohésion effective et l’angle de résistance, et de discuter le fluage d’un sol drainé pour différentes conditions de chargement. Le Président :

Thank you, Mr Meyerhof. I will now ask the panel to open the discussion of subject (a). Dr Bishop, would you be so kind as to open the discussion? M. B ishop (Grande-Bretagne) In my contribution to this discussion I want to consider the problem facing the engineer; whether to accept the scatter of his test results and analyse them statistically, drawing sometimes rather pessimistic conclusions, or whether to track down the source of the scatter and see if it does not in fact often lie in the testing technique rather than in the natural properties of the soil. There are two particular sources of error to which I wish to refer. The first arises from the fact that in undrained tri­ axial compression tests run at conventional rates of testing there may be large differences in pore water pressure between the middle and ends of the specimen. This is partly a con­ sequence of the restraint caused by the rigid plattens at the ends of the specimen and partly a function of the properties of the soil itself. We were aware that this difference might occur, but in recent years we have found that its magnitude is much greater than anticipated and can lead to a very serious error in soil properties based on a single point pore pressure measurement unless the test is run sufficiently slowly for equil­

ibrium to be set up within the pore water throughout the whole specimen. This may be illustrated by test results presented at the recent Conference on the Shear Strength of Cohesive Soils held by the American Society of Civil Engineers. Fig. 1 shows the difference between the pore water pressure measured at the base of the specimen and that measured at its mid-height in a test of 8 hours duration on a sample of compacted clay shale 8 inches in height. The difference is of the order of 10 lb./sq. in. at failure at a rate considered to be on the conser­ vative side in ordinary testing practice. In Fig. 2 the corresponding results of a test on the same soil of about 120 hours duration are illustrated. The differ­ ence between the base and mid-height values of pore water pressure has decreased to about 2 lb./sq. in. which can be ignored without serious error. The effect of unequalized pore pressures on the position of the failure envelope is illustrated in Fig. 3. The base measure­ ment of pore water pressure in the 8 hours test leads to a cohesion intercept of 11 lb./sq. in., while the mid-height measurement indicates an intercept of only about 1 lb./sq. in. This difference has very serious implications from a design point of view, and in many of the test results still being publised this factor is overlooked. In Fig. 4 the difference in pore water pressure between the middle and ends of the specimen is plotted against the time to reach 10 per cent axial strain. Given sufficient time the pore pressure will equalize and a valid result can be obtain­ ed from a single point measurement. The time required, however, for this particular soil (22 per cent clay fraction) and sample height (8 inches) is about 10,000 minutes or 7 days, unless special means are taken to accelerate the equal­ ization of pore pressure. Strips of filter paper around the

STRAIN /. o Z

o

IO

O O;

I5

-os Ü-1-0 z

Fig. 1 Effect of rate of strain on pore pressures measured at end and centre of a Rate of strain: 20 % in 8 hours.

A"

dia x 8" high sample of compacted shale.

89

Fig. 2 Effect of rate of strain on pore pressures measured at end and centre of a 4" dia X 8'' high sample of compacted shale. Rate of strain: 20% in 120 hours.

perimeter of the sample may be used for this purpose in saturated soils. Fig. 5 shows the water content profiles of two typical samples of the compacted clay shale, taken after time had been allowed for the pore water pressure gradients to approach zero. The profiles clearly reflect the observed differences in pore water pressure. For different types of soil this difference may be a differ­ ence not only in magnitude but also in sign. For normally consolidated sensitive clays the pore pressure is higher in the middle of the sample than at the ends. This is illustrated by the results of tests by Taylor and Clough (1951) on un­

O

IO

20 EFFECTIVE

30 NORMAL

disturbed Boston clay, one of which is plotted in Fig. 6. Also plotted are the results of a test on a saturated remoulded clay, which is lightly over-consolidated. Here the pore pressure difference is small. In contrast the compacted clay shale shows a large difference, the pore water pressure being lower in the middle than at the ends. A theoretical relationship has been worked out by Dr R.E. Gibson between the duration of the test, expressed as a dimensionless time factor, and the degree of equalization of the initially non-uniform pore pressure. For a given height of sample and coefficient of consolidation the appropriate testing time can be selected. This relationship is plotted in

40 STRESS

50

60

70

PS.I.

Fig. 3 Effect of unequalized pore pressures on apparent position of failure envelope for compacted shale. Rate of strain: 20 % in 8 hours.

90

80

Q >

IO

IOO

IOOO TI ME

TO

R E A CH

107.

lOOOO

IOOOOO

S T R A I N — MI N.

Fig. 4 Effect of time of testing on equalization of pore pressure during shear of compacted clay shale.

Fig. 5 Equilibrium water content profiles.

Fig. 7, together with the results of tests carried out at Imperial College and at M.I.T., which are in reasonable conformity with it. The second error to which I wish to refer arises in partly, saturated soil from the difficulty in distinguishing between the pore water pressure and the pore air pressure. In Fig. 8 the results are plotted of some triaxial tests in which fine ceramic discs of high air entry value were used to measure the pore water pressure, the pore air pressure being measured separately with discs of coarse woven fibre glass cloth at the opposite end of the sample. The difference between the pore water pressure and the pore air pressure due to surface tension

varies from zero to nearly 20 lb./sq. inch in the series of tests illustrated. If the strength is plotted against effective stress on the basis of the difference between total stress and pore water pressure the relationship shown by square symbols is obtained. This line, if extended, would give a negative cohesion intercept. If the results are plotted in terms of the difference between total stress and pore air pressure the relationship shown by round symbols is obtained. This corresponds to a large positive cohesion intercept. Tests on fully saturated samples in which there is no am­ biguity about the determination of effective stress, lead to

91

STRAIN 7.

Fig. 6 Difference between centre and end pore pressure measurement for three soils representing a wide range of « A » values. an intermediate position for the strength plot, with a small positive cohesion intercept. Since in the partly saturated soil neither the air pressure nor the water pressure can in general act over the whole of the surface of the soil particles, the use of one or the other alone will lead to an incorrect estimate of effective stress. This has led to the proposal of a modified effective stress equation :

This equation is discussed in a paper to this Conference (Bishop and Donald, 1961) and a more general examination of the problem is to be found in the Proceedings of the Confer­ ence on Pore Pressure and Suction in Soil, London. 1960. If the additional factor controlling effective stress in partly saturated soil is ignored, apparent inconsistencies may arise between the results of different types of test on the same sample. In particular, if coarse grained porous discs of low air entry value are used, pore air pressure will usually be ° — "a + 1 (“a — “J measured, and this, as shown by Fig. 8 will lead to a very high apparent cohesion intercept, which is not a true property where a' denotes effective stress, cr — total stress, of the soil. If the porous disc, being initially fully saturated, ua — pore air pressure, sometimes measures pore water pressure, sometimes pore air pressure, considerable scatter of the results may result. uw — pore water pressure, and •/ — a parameter varying between 1 for fully I have drawn attention to these two sources of error, saturated soils and 0 for dry soils, because very inconsistent results may be obtained if they depending mainly on the degree of are not given full consideration, and the application of statis­ saturation, but also on soil type and tical methods will not tell you which is the right result cycle of wetting or drying, etc. to apply in any particular practical problem. 92

}

T FROM CONS T ANT

FI RST

RATE

y

# "1

■

T FROM

j

S T A GE

OF

E QUI V A L E NT

REDUCED

RATE

STRAI N

SECOND

( lO°/J

STAGE

H2

Fig’ 1 Comparison of test results with theoretical relationship between percentage equalization of pore pressure in undrained tests and time factor. Compacted clay shale at two water contents and Taylor’s tests on undisturbed Boston clay.

074-03 — U w 6 2

0 7 + 03 — u a

at

failure

psi

2

Fig. 8 Triaxial tests on a compacted shale (clay fraction 22 %) compacted at a water content of 18-6% and sheared at constant water content.

Références [1]

G.E., and D o n a l d , I.B . (1960), " Factors controlling the strength of partly saturated cohesive soils ”, Proc. Conf. on Shear Strength o f Cohesive Soils, Amer. Soc. Civ. Engrs. ; pp. 503-532 [2] —, B l i g h t , G.E. and D o n a l d , I.B. (I960). Discussion. Proc. Conf. on Shear Strength o f Cohesive Soils, Amer. Soc. Civ. Engrs.; pp. 1 027-1 042. [3] — , and D o n a l d , I.B . (1961) " The experimental study of partly saturated soil in the triaxial apparatus ”. Proc. 5th Int. Conf. Soil Mech., vol. 1 ; pp. 13-21. [4] Proc. Conf. on Pore Pressure and Suction in Soils. Butterworths, London, 1960. [5] T a y l o r , D .W . and C l o u g h , R.H. (1951). Research on shearing characteristics o f day. M.I.T. Report. B is h o p , A .W . A l p a n , I ., B l i g h t ,

M . M archand (France).

Je serai tenté d’abord de répondre à M. Bishop, car son exposé m ’a vivement intéressé, ayant eu récemment quelques difficultés à interpréter certains essais de mesure de pression interstitielle effectués sur des matériaux grossiers, non saturés, au cours de cisaillements à l’appareil triaxial. Je crois que l’on pourrait tirer deux conclusions de son exposé, peut-être un peu extrêmes : — D ’une part pour faire des mesures de pression intersti­ tielle valables, il est nécessaire de faire des essais assez lents. Dans ce cas au lieu d’essais non drainés et lents, on pourrait aussi bien faire des essais drainés qui se passent de mesures de pression interstitielles, au moins dans de nombreux cas 93

où l'intérêt des essais non drainés avec mesure des pressions interstitielles réside dans la rapidité d’exécution de l’essai. — D ’autre part, il semble que lorsqu’on augmente la pression dans la chambre triaxiale, les différences entre les pressions de l’air et les pressions de l’eau diminuent, du moins les différences relatives, d’où l’intérêt des essais à très forte charge pour déterminer avec un minimum d’erreur l’angle de frottement interne des matériaux non saturés. Je voudrais passer maintenant plus spécialement à une question de dispersion des essais de mécanique des sols. Je crois qu’il y a beaucoup à dire là-dessus. La dispersion com­ mence à l’échantillonnage, suit dans l’essai proprement dit et se termine dans son interprétation. Tous les essais n’ont pas la même importance. J’ai l’im­ pression que pour les essais de classement, on peut se conten­ ter d’une large dispersion étant donné que l’on en attend surtout des renseignements qualitatifs. On pourrait en dire autant des essais de compactage et de tassement parce que l’on observe fréquemment des différences suffisamment sen­ sibles entre le laboratoire et le chantier pour que l’on n’exige Fig. 9 Cisaillements triaxiaux non consolidés et non drainés pas de ces essais une grande précision. normaux et cell-test. Matériaux 0,20 mm compensés. Par contre, l’essai de cisaillement, lui, requiert la plus grande Contrainte sur le plan de rupture en fin d ’essai. précision, car on l’utilise directement dans des études de stabilité, souvent avec des coefficients de sécurité très réduits qui ne tolèrent pas une grande marge d’erreur. 31° et 39°; quand on porte tous les résultats sur le même Pour l’utilisateur que je représente, n’ayant pas person­ graphique, on s’aperçoit que l’on peut retenir en fait un angle nellement de laboratoire, le premier problème consiste à de 37° avec une bonne précision, ainsi qu’en témoigne la faire la part entre la dispersion inévitable et l’erreur. Actuelle­ figure 9; une petite erreur sur la pression interstitielle, par ment je crois qu’il n’y a pas d’autre méthode pour y arriver exemple attribuable au défaut de saturation, a vite fait de que de multiplier les prélèvements, les essais, les méthodes faire basculer la droite de Coulomb, surtout si l’on travaille d’essais, et même, lorsque c’est possible, les laboratoires, dans un domaine de pression restreint. A défaut d ’un grand pour essayer de diminuer l’influence du matériel et du personnel. nombre d’essais on peut toujours diminuer les risques de Il sera certainement possible de diminuer ce nombre d’essais l’interprétation individuelle en précisant bien le domaine nécessaires lorsqu’auront été établies les cartes géotechniques d’emploi de la formule en cohésion et frottement. dont M. Meyerhof parle dans son rapport général, cartes Dans le but de réduire cette dispersion due aux éprouvettes, qui permettront de préciser certains rapports entre les essais tout en cherchant à réduire les manipulations dans le cas de de classement et les essais mécaniques. Naturellement, cela matériaux compactés et de cisaillement directs, nous avons ne dispensera pas de faire des essais mécaniques, mais cela utilisé un moule Proctor spécial que l’on pouvait diviser guidera l’opérateur et lui montrera tout de suite si le résultat en un certain nombre de boîtes. Je crois que l’on avait ainsi obtenu est normal ou non, le résultat anormal pouvant le maximum de chances d ’obtenir des éprouvettes analogues, d’ailleurs être valable. Dans le premier cas on se contentera à condition d’éliminer les extrémités du moulage. de quelques essais de contrôle, dans le second il faudra faire Il y a aussi une méthode intéressante, la méthode hollan­ une étude complète et-préciser ou corriger la carte. daise du Cell-test, qui se contente d’une seule éprouvette. Toutefois, il y a une difficulté particulière pour l’essai de Elle permet de gagner du temps, mais en contrepartie elle cisaillement dans ces cartes géotechniques, parce que les exige une grande dextérité de l’opérateur pour ne pas trop appareillages varient beaucoup d’un laboratoire à l’autre; déformer l’éprouvette au premier chargement. ils sont souvent de conception originale; les méthodes d’essai Pour ces essais systématiques ayant partiellement pour aussi. Donc, je me demande s’il ne serait pas intéressant de but d ’estimer la dispersion, on peut souvent se contenter normaliser certains types d’essais de cisaillement, ce qui d ’essais un peu simplifiés. Ainsi nous utilisons assez couram­ n ’empêcherait pas d’ailleurs de faire des essais non normalisés ment en France, dans les cas des sols grossiers, la méthode et au besoin de changer les normes si l’on s’aperçoit qu’elles de compensation pour préparer le matériau des éprouvettes sont dépassées par la technique. compactées; elle consiste à remplacer les gros éléments par De toutes façons le projeteur doit rester extrêmement des éléments plus petits en poids égal. Par exemple, pour re­ méfiant devant les progrès de la technique; par exemple présenter un matériau naturel 0-200 mm, nous avons fabriqué si un essai nouveau, indiscutablement correct, montre que les des matériaux 0-60 mm où la partie 60-200 mm du matériau résistances au cisaillement étaient jusque là sous estimées, naturel était remplacée par du 20-60 mm, des matériaux cela ne veut pas dire qu’il faille aussitôt raidir les pentes des 0-20 mm où la partie 20-200 mm était remplacée par du barrages en terre en projet. Il faut peut-être corrélativement 5-20 mm. Et même, nous avons été, pour tester la méthode, changer le coefficient de sécurité du calcul de stabilité. jusqu’à fabriquer du 0-5 mm, où la partie 5-200 mm était En ce qui concerne le cisaillement, j ’ai l’impression qu’une remplacée par du 2-5 mm. Il faut vous dire que la partie des causes de dispersion les plus nettes réside dans les diverses 5-200 mm représentait 60 pour cent du matériau, c’était donc natures des éprouvettes que l’on cisaille. On utilise, en général, une compensation extrêmement poussée. Néanmoins, on a trois éprouvettes, et l’on sait qu’elles ne sont jamais absolu­ obtenu des résultats valables, même avec ce 0-5 mm, comme ment identiques. Cela peut conduire à des erreurs sensibles on peut le voir sur la figure 10. Tout ce que l’on a constaté, si, considérant isolément un essai sur trois éprouvettes, on c’est qu’au compactage la granulométrie subissait un change­ l’interprète individuellement par une cohésion et un frotte­ ment notable, mais c’est un autre phénomène indépendant ment auxquels on est tenté de conférer une réalité physique de la méthode de compensation. indépendante du domaine restreint de l’essai. Par exemple, Cette méthode de compensation n ’est d ’ailleurs qu’excep­ sur soixante essais, on obtient par cette méthode d’interpré­ tionnellement utilisée pour des essais de cisaillement car il tation individuelle des angles de frottement qui varient entre est bien rare que la granulométrie des terres le justifie. Par

M a té ria u * compense’s

Elisais non cony>licc.'m height 7,60 fn,n

ed

{00mm

S o il

0 -2 0 mm '« ilk c o m p e n s a tio n .

Fig. 11 Effet de paroi. contre son emploi pour le compactage est plus courant. Des essais comparatifs, en compactage et perméabilité, portant sur des matériaux compensés 0-60 mm placés dans des moules de 400 mm de diamètre et des matériaux compensés 0-20 mm placés dans des moules CBR de 150 mm de diamètre, ont montré que les résultats différaient suffisamment peu pour que soit justifiée la validité de la méthode. Dans le même esprit de simplification, je crois que l’on peut essayer de s’affranchir de certaines règles habituelles surtout lorsque ces dernières conduiraient à renoncer à l’essai, un essai imparfait étant préférable à pas d’essai du tout. Ainsi, dans un cisaillement direct avec des boîtes telles que l’épaisseur totale de l’échantillon soit de 50 mm nous avons utilisé des matériaux compensés avec des éléments allant jusqu’à 20 mm, ce qui, à priori, paraissait extrêmement abusif. Néanmoins, la figure 11 montre que le résultat est très correct et en très bonne concordance avec l’essai triaxial correspondant, sur éprouvette de 100 mm de diamètre et 250 mm de haut. Finalement, en travaillant dans les meilleures conditions possibles, je crois que l’on devrait arriver, sur l’essai de cisaillement drainé, à une précision de l’ordre de 1 ou 2°, ce qui est amplement suffisant si l’on considère que le matériau naturel ou fabriqué varie toujours assez fortement d’un point à un autre; on est donc obligé de tenir compte d’un matériau sinon moyen du moins limite du côté de la sécurité et la dis­ persion propre de l’essai devient assez négligeable. Néanmoins, il convient que les essais soient parfaitement exécutés, et que les normes soient toujours bien respectées. En particulier, il faut se méfier de la routine qui tend à faire dévier tout doucement des normes, ce qui peut conduire par­ fois à des résultats tout à fait scabreux. En conclusion je vous prierai de bien observer que mon propos était essentiellement pratique et n’avait nullement la prétention d’apporter des précisions aux remarquables études récentes sur le cisaillement des sols, en particulier à celles présentées il y a un an à la Conférence de l’Université du Colorado.

Le Président : Thank you very much, Mr Marchand. Well, that concludes the panel’s discussion on subject (a). We now go on to subject (6), the dissipation of pore pressures — and I will ask Mr Denissov if he will be so kind as to open the discussion. N.

(U.R.S.S.) I should like to say a little about overestimation of the influence of pore pressure on clay strength in the problem of the stability of clay slopes. The problem of pore pressure dissipation is only problem (b) in the list of subjects for discussion but I think this is a very important question — in fact, it is problem No. 1 in modern soil mechanics. At the present time much attention is being paid to the effect of pore pressure upon the strength of soils. Of special interest are the well-known papers by Bishop, Skempton and a number of other scientists and engineers. The problem of the pore pressure increase's influence upon soil strength is, generally speaking, that of their dependence on pressure. And the problem of the pore pressure increase’s effect on clay slope stabilities is reduced to that of the poss­ ibility and degree of clay strength drop as the result of their unloading. For sand, the strength of which is solely due to the effect of internal friction, the influence of pore pressure is surely and easily determined by the simplest experiments. When, however, sand is transformed into sandstone, which is accompanied by the development of stiff ” cementation bonds, the possibility of the effect of pore pressure upon the strength is eliminated. It is obvious that pore pressure can neither influence the strength of cementation bonds nor the strength of the particles. The state of sandstone in natural conditions may be considered as near that of unjacketed samples in the labor­ atory. The experiments made with sandstone, marbles and concretes carried out by different researches (A.W. Skempton, D

e n is s o v

95

1961) show that pore pressure increase under such conditions does not cause any drop in the strength of materials. In recent years many papers have been published (Ivan Popov, U.S.S.R., 1941, V. Florin 1959, N. Denissov 1941 and 1956, N. Denissov and P. Rebinder, 1946, T.W. Lamb 1961, L. Bjerrum and T.H. Wu, 1960, 1. Rosenquist, 1959, H. Seed, J, Mitchell and C. Chan, 1960), in which more and more attention has been paid to the concept that the strength of clay is to a great extent due to the existence of stiff bond between the particles, developing because of the cementation of soil. The effect of the increase of pore pressure on strength can be manifest here as well as for sandstone only after the collapse of this bond. Just by the effect of cementation bond and its elimination can be explained the well-known difference in the strength of natural clay and remoulded samples of equal density. Now let us consider the clays having no cementing bonds; their cohesion results from the influence of Van-der-Vals’s forces, etc. It is well known that both the drained strength of these soils and their density increase under the growing pressure. The pore pressure developing under the place of contact of the soil and the loaded surface somewhat diminishes the effective influence of the consolidating pressure. This, of course, tells on the rate of soil strength increment. Pore pressure dissipation in time is accompanied by a greater influence of an external load and by a higher degree of soil consolidation and its strength. When a decreasing pressure acts upon such soils, their density and strength remain practically unchanged. To prove this one can refer to the results of laboratory investigations and field observations which show the con­ siderable strength of over-consolidated clays. It can be explained that in the process of previous consolidation their strength grew mainly because of bond increment accompanying the rise of particle concentration in a unit volume. The above­ said may be illustrated by the test results for shear resistance of the alluvial clay (Fig. 12) in the normally consolidated state (line 1) when the density changed with pressure increase, and in the over-consolidated state (line 2) when the density change resulted from pressure decrease. Z

F ig . 12

where Cmin — initial cohesion a — pressure. For unloading T0" = Cmin + CTtg ^ — A a tg A a — pressure decrease. From the above-stated, it follows that the pore pressure increment in the non-cemented clays forming the slope and diminishing effective stresses cannot substantially influence the strength of the clays and the stability of the slope. At the same time it is clear that the use of the ordinary value of the angle o f friction, the pore pressure action on the stability of slopes being taken into consideration, leads to a considerable overestimating of this influence. Thus the shear strength of the alluvial clay (Fig. 12) under the pressure of 4 kg./cm2 is 1-73 kg./cm2. The pore pressure increase by 2 kg./cm2 leads to a corresponding decrease of effective stress. If the shear resistance, corresponding to this new stress, is determined from the straight line 2 we receive 1-62 kg./cm2, and from the straight line 1 only 1-06 kg./cm2. Thus with correct estimate of pore pressure influence the strength of the clay decreased by 6 per cent, and when using the relation illustrated by the straight line 1, the strength of the clays decreased by 40 per cent. This can explain the fact that the coefficients of stability for a number of calculated slopes are much less than one. It is necessary to note that, as opposed to sands, the shear resistance of clays can change with the stress alternating only in cases when the alterations of this state is accompanied by the change of density and structure. Pore pressure fluc­ tuations in the marine clays which are, in the slopes, in an over-consolidated state, cannot be accompanied by the volume change of these clays. That is why their strength can decrease in time as a result of the physical and chemical influence of pore water but not of its mechanical action. According to the above-stated it can be said that the effect of pore pressure undoubtedly influences the strength of sand. The effect of this pressure upon the strength of clay soils with natural structure is usually estimated on the basis of laboratory tests on jacketed samples. No doubt, there are natural conditions corresponding to these laboratory tests when a layer of a water-saturated soil is between the strata of practically impervious soils. How­ ever, the state of clay strata similar in their permiability somewhat differs from these conditions. One thinks that the difference in permeability is the only reason for the distinction of properties of clay and sand. This explanation is not enough. The main reason for it is the different degree of colloidal-chemical influence in the strength of clay and sand. From this point of view, because of the usual ignoring of both the residual nature of the strength of clay and the gain in strength of clay soils and the influence of cementation bonds, the effect of pore water pressure is substantially over­ estimated. Whilst expressing a doubt about the possibility of the effective influence of pore water pressure on the strength of clay soils, it is at the same time impossible to neglect the great importance of observation of the rate of pore pressure dissipation. These observations give reliable data which enables one to judge the extent of the completeness of the process of the compression or expansion of soils and the rate of these processes. This in turn enables one to adjust the rate for construction work.

The experiments were carried out under drainage and the stresses at the abscissa axes are effective. The figure clearly shows that the angle typical of the soil strength increase under loading, is substantially larger than the angle di2, typical of strength drop under unloading. Hence, for different schemes of stress changes we must use different expressions showing relations between shear resistance t and consolid­ Références ating pressure a : [1] B , L. and T H W (1960). “ Fundamental For loading Shear-Strength Properties of the Lilia Edet Clay ”, GeoCmin O tg technique, No. 3. je r r u m

T 0 '

96

=

+

ie n

s in g

u

[2] D enissov , N. (1941). " Causes Underlying the Coherence of Loess-like L oam s”. Comptes rendus (Doktady) de VAcadémie des Sciences de l'U.R.S.S., vol. XXXII, No. 4. [3] — (1956). “ Engineering Properties of Clay ”. Mos­ cow, Gosenergoizdat. [4] —, and R ebinder P. (1946). “ On Colloid Chemical Nature Cohesion in Argillaceous Rocks. ” Comptes Rendus (Doklady) de VAcadémie des Sciences de /’ U.R.S.S., vol. LIY, No. 6. [5] F lo rin , V. (1959). Soil Mechanics, vol. I, Leningrad. [6] L ambe T.W. (1960). " A Mechanistic Picture of Shear Strength in Clay. ” Report for the ASCE Research Confe­ rence on Shear Strength of Cohesive Soils, Boulder, Colo­ rado. [7] Popov, I. (1941). Laboratory's Investigations of Soils, Moscow, 1941. [8] R osenqvist I. (1959). " Physico-chemical Properties of Soils : Soil Water Systems ”. American Society of Civil Engineers. Proceedings, vol. 85, SM. 2. [9] S eed H.B., M itchel J.K., C han C.K. (1960). " The Strength of Compactes Cohesive Soils Report for ASCE Research Conference on Shear Strength of Cohesive Soils, Boulder, Colorado. [10] S kempton A.W. (1960). " Effective Stress in Soils, Con­ crete and Rocks, Pore Pressure and Suction in Soils ”, London, 1961.

Le Président : Thank you very much, Dr Denissov. Would Dr Bishop care to continue the discussion? M. B ishop I will try and deal first with one of the points which Prof. Denissov has just raised. I have the impression that the cemented bonds to which Prof. Denissov refers have not been wholly over-looked in previous work relating to the principle of effective stress. Work which we carried out some years ago (Bishop and Eldin, 1950) on the influence of a finite contact area between the particles showed that the difference between the total stress g and the pore pressure u controlled the deformation properties when the contact area was large just as it does in an uncemented sand at ordinary stresses where the contact area is extremely small. Tests on concrete by the U.S. Bureau of Reclamation (McHenry, 1948) and by other workers indicate that the strength of such materials, at least at com­

plete rupture, is controlled, to a close approximation, by the stress difference a — u. More recent tests in Portugal by Serafim (1954) show that the principle of effective stress applies also to the deformation of cemented materials, although account has to be taken of the behaviour of the material forming the bond. Since time is short I could not do better than refer to the introductory address given by Prof. Skempton to the Confer­ ence on Pore Pressure and Suction in Soils in London (1960), in which much of this data is summarised. Theories were also presented in this address which apply to both sands and similar materials with an extremely small contact area and to concrete and natural sandstone and other rocks where the cemented area is very large. It appeared from Prof. Skempton’s investigation that there is no discontinuity in the applic­ ation of the principle of effective stress. As far as I can understand from the previous discussion, the suggestion that pore pressure has no influence on strength, in the presence of stiff cementation bonds is a theoretical concept not yet supported by experimental evidence. I would agree that more attention needs to be given to the behaviour of natural samples at very small strains, but field evidence as well as laboratory tests will be needed before we reject a principle which hitherto has been extremely useful to us in understanding the properties of soil. My second point relates to the dissipation of pore pressure in the field. During the construction of the Selset dam in the north of England the opportunity was taken of confirming the design assumptions by a detailed series of field measure­ ments of pore pressure. There were two aspects to the problem. The dam itself was built on a soft boulder clay foundation which it was uneconomical to strip, and 18 inch diameter sand drains at 10 ft. centres were used to accelerate the consolidation of the foundation. Piezometers were placed at 14 points in the middle of groups of these sand drains to measure the dissipation of pore pressure and thus enable the field value of coefficient of consolidation to be compared with the labor­ atory value. In addition the fill was to consists of the same impervious clay, having a permeability of about 1 x 10-8 cm. per sec. The average rainfall was expected to be about 20 inches in each construction season. The control of water content necessary to give low pore water pressures would have been uneconomical under these circumstances. The embankment

STONE PITCHING PUDDLE

ROLLED BOULDER CLAY FILL

C l AY

CRUSHED STONE

DRAINAGE BLANKETS

BEACHING

HEIGHT OF DAM 125'

18' CXA. SAND DRAINS AT IO ' CENTRES DRIVEN THROUGH A LLU V IAL GRAVEL AND SOFT BOULDER CLAY

SOFT AT TOP

GRAVEL

SHALE

CONCRETE C U T -O FF 6 ' WIDE

GROUT CURTAIN BELOW CONCRETE C U T -O F F

SCALE

Fig. 13 Selset Reservoir : Typical cross-section of embankment.

97

2Bo

Fig. 14 Observations of pore water pressure in the clay foundation.

was therefore divided into layers sandwiched between horiz­ ontal drainage blankets at 15 ft. centres to permit rapid consolidation of the fill. In the lower part of the dam vertical as well as horizontal drains were used to take advantage of the horizontal permeability if it proved to be greater than that in a vertical direction. The cross section is shown in Fig. 13. Fuller details of the construction, instrumentation and analysis of the results will be found in other papers (Bishop, Kennard and Penman, 1960; Kennard and Kennard, 1962; Bishop and Vaughan, 1962). The observations are at present being analysed in detail, but the preliminary results indicate that the foundation pore pressures have dissipated a little faster than predicted. Fig. 14 shows the build-up of load on the foundation and the corres­ ponding changes in pore water pressure. The rapid increase in pore pressure during the construction season and the decrease during the shut-down period are most marked. The preliminary estimate of coefficient of consolidation cv from the field results gives a value approximately three times the laboratory value based on 4 inch diameter samples. In selecting the samples in the trial pits we obviously had to avoid the most stoney zones of the boulder clay, and this would have weighted the average of the laboratory tests to give a lower value of cv than would be obtained from fully representative samples. However, considering the possible errors in measurement and analysis, I think the agreement is not unsatisfactory, particularly as the field value lies on the right side from the construction point of view. In the compacted fill, which was mainly placed wet of the optimum water content, the average field value of the coefficient of consolidation as determined from the piezo­ meter results is about 1-5 times the laboratory value based on samples from which material greater than 3/8 inch dia­ meter was excluded. Within the fill itself material up to about 1 ft. diameter was included. The agreement is satisfactory — the error is again on the safe side. These results show that we can rely on dissipation of pore 98

pressure in field problems and that we can make valid pre­ dictions on the basis of laboratory tests provided we use a certain amount of caution in selecting representative values. My third point is that the final value of pore pressure depends not only on the rate of dissipation but also on the initial value of the pore pressure, and this depends on the magnitudes of the three principal stresses in the field. The General Reporter has drawn attention to the lack of data about the behaviour of soil under plane strain conditions, which apply in many engineering problems. I will give a few illustrations, therefore, of the kind of results we have been obtaining with the plane strain apparatus at Imperial College. D r Clive Wood, Dr Cornforth and I hope to publish the detailed results in the near future.

VALUES

AT M A X I M U M

PRI NCI PAL

ST RESS

RATIO

°v

—L

Fig. 15 The relation between strength and effective normal stress; a comparison of the results of plane strain and cylindrical compression tests on compacted soil.

«✓ »

Fig. 16 The relation between pore pressure and major principal stress in undrained tests: a comparison between plane strain and cylindrical compression tests.

The apparatus will be familiar to those of you who visited our laboratory at the time of the London Conference. Sam­ ples 4 inches in height, 2 inches in thickness and 16 inches in length are tested in a cell in which the intermediate prin­ cipal stress is measured under the condition of zero strain in that direction. Tests have been completed on computed fill and cohesionless sand and are at present in progress on a saturated clay. Fig. 15 shows the relation between strength and effective normal stress for the compacted soil. A higher strength is clearly obtained in plane strain, the difference in the tangent of the angle of shear resistance 0 ' being about 10 per cent. This is significant from the engineering point of view although not dramatic. The difference is on the safe side if our estimates of factor of safety are based on cylindrical compression tests. However Fig. 16 shows that the pore pressures set up under plane strain conditions are higher than those obtained in the conventional cylindrical compression tests, both at the maximum principal stress ratio and at the maximum deviator stress. Field values of pore pressure may therefore consid­ erably exceed estimates based on the results of cylindrical compression tests, though if provision has been made for their observation and control this situation should be revealed and met without difficulty. The tests on sand, Fig. 17, show the same trend towards a higher strength in plane strain in the denser samples. In the loose samples the difference is small. The trend towards a higher strength is accompanied by a smaller strain at failure and a smaller increase in volume, in the drained tests, before failure occurs. This corresponds to the higher pore pressures observed in the undrained plane strain tests. The difference in the angle of internal friction of about 4° in the case of dense sand would imply an increase of 10 to 20 per cent in the factor of safety, in a slope stability problem,

over that based on cylindrical compression tests. In terms of bearing capacity the implied difference is very much greater. In Fig. 18 the increase in bearing capacity of a strip footing if the estimate is based on the plane strain test rather than the cylindrical compression test is plotted against porosity. The difference ranges between 30 per cent and 13 per cent. It is an interesting reflection that many people have obtained ■ PLANE .STRfW □ PtBME 3TRglM WITHOUT EMO CtgffP • 5TBMPBRP TRlffXIBL

46°

a« & Uol 43 "T3 4a xx UJ

■ \B ■ \\■ nsr 150 4140 » \ ■r & 39 •V

I-»

i 36

3*

g - 33 32

34

35

36

37

33

39

40

41 42 43 44 ft CM Fig. 17 The relation between drained angle of internal friction and initial porosity: a comparison between plane strain and cylindrical compression tests on sand. WERgGE IMITIflL POIjOSITY

99

34

35

37 38 39 40 ftfERtyGE IKlTlgl POROSITY ni. (%)

36

41

42

43

Fig. 18 The influence of type of test on calculated bearing capacity.

agreement between a proposed theory and the results of bearing capacity tests in terms of the cylindrical compression test. I hope they can now provide an equally rational explan­ ation in terms of the plane strain test! Références [1] [2] [3] [4] [5] [6] [7]

A,W, and E l d in , A.K.G. (1950). " UndraLned Triaxial Tests on Saturated Sands and their Significance in the General Theory of Shear Strength ”. Geotechnique, vol. 2; p. 13. —, K e n n a r d M.F. and P e n m a n , A.D.M. (1960). " Pore Pressure Observations at Selset Dam ”. Proc. Conf. on Pore Pressure and Suction in Soils, p. 91, Butterworths, London. —, and V a u g h a n , P.R. (1962). “ Selset Reservoir : Design and Performance of the Embankment Proc. Instil. C ivil Engrs, No. 6 589. K e n n a r d , J. and K e n n a r d , M.F. (1962). “The Design and Construction of Selset Reservoir ” Proc. Inst. C ivil Engrs, vol. 21, Febr. 1962; p. 279. M e H e n r y , D. (1948). " The Effect of Uplift Pressure on the Shearing Strength of Concrete ”. Proc. 3rd Congr. Large Dams. S e r a f im , J.B. (1954). " A Subrasao nas Barragens ”, M in. das Ovras Publi., Lisbon. S k e m p t o n , A.W. (1960). " Effective Stress in S o ils, C o n ­ crete and Rocks ”, Proc. Conf. on Pore Pressure and Suction in Soils, p. 4, Butterworths, London.

B is h o p ,

Le Président : Thank you, Dr Bishop. Would Dr Breth care to make a comment? M. H. B r e t h (Allemagne) Mr Bishop’s most interesting investigation shows us that the air pressure and the water pressure in the pore space of partly saturated soils are different. One can expect that those pressures will be in a state of equilibrium if the soil is consol­ idated. In this state air pressure and water pressure in the pore space should be equal. If the air cannot get out of the pore space, or only after a long time, a residual pore pressure should occur. During drained tests we several times measured a residual pore pressure. After what Mr Bishop said several problems 100

result for the praxis. With the usual measuring system we receive only the water pressure or the water pressure influenced by the air pressure. The result is that we do not obtain any regularity. We have measured pore water pressure in partly saturated soils in the center of large samples by triaxial tests and found out that pore water pressure depended upon the dimension and the rate of the load increment. The influences changed with the preparation and the compaction of the sample and with the state of consolidation. For instance, in a blanket of clay of 6 m. thickness under a dam, pore pressure due to quick drawdowns showed greater change during the first year than during the following years. Again we can see the influence of time. Thus it will be very difficult to draw conclusions from the results of investigation into actual problems of engineering construction. Le President : Thank you, Mr Breth. I would like to introduce Prof. Geuze to this discussion. I think you know Mr Tan Tjong Kie was originally a member of the panel. He is unfortunately unable to attend and Prof. Geuze at very short notice has been so kind as to join us. Would you like to make some comments on section (b), Prof. Geuze? M. E. G e u z e (Etats-Unis) Thank you Mr Chairman, I would like to do that very much. May I use the privilege of continuing my remarks on question (c) after those I will make in connection with ques­ tion (b). My reason for this is that the phenomenon of pore pressure and its dissipation obviously depends on the differ­ ence between the elastic and the viscous properties of the soil. These properties are in fact so closely interrelated, that we could hardly discuss either one of them without reference to the other. Magnitudes of pore pressure in the voids of a supposedly saturated soil system depend to a great extent on the nature of the forces acting between the particles of the system. In the Terzaghi theory of consolidation these forces are assumed to be in the nature of a one-to-one relationship between the stresses and the strains. Considering the comparitively ideal mechanical properties of the pore fluid the relationship bet­ ween the effective and neutral part of the stress is therefore a simple one. If however the nature of the stress-strain relationship is complicated by a deviation from the assumed linear, Hookean behaviour of the soil skeleton, we can only assess its effect on the bulk behaviour indirectly through the measurement of pore pressure. I understand from Dr Denissov that he has considered this effect in dealing with soil systems subjected to cement­ ation bonds between the particles. The properties of the soil structure then become of primary importance and I actually consider this to be our central problem in the study of the mechanical properties of saturated clays. Since we are obviously not in a position to remove the fluid from the voids without affecting the magnitude (and sometimes even the nature) of the particle bonds, we can only assess the effects of the nature of these bonds on the behaviour of the soil structure by indirect means, i.e. by the measurement of the pore pressures in the voids of the soil system (which we normally do when measuring on the gravity scale) and the stress-strain (both hydrostatic and deviatoric components) relations of the bulk material. This problem appears when we consider the rate of pore pressure dissipation in elasto-viscous soil systems. The classic concept of Terzaghi’s consolidation theory is then complic­

ated by the fact that the soil particle structure will show time-dependent deformations in shear and compression, which can not be explained by a simple elastic stress-strain model of the soil structure. In other words the Terzaghi theory of consolidation cannot be applied to soil systems with elasto-viscous properties. Successful attempts have been made over the last seven years to use simplified Theolog­ ical models of elasto-viscous soil systems in order to estab­ lish their mechanical properties both in consolidation and in stress-strain behaviour of saturated clay systems (Tan Tjong Kie 1954, Schiffman 1959, Gibson 1960). These simpli­ fications mostly bear on the elastic nature of the hydrostatic stress-strain relationship and the unique stress-rate of strain relationship of the deviatoric part of the stress system. Both the bulk modulus and the viscosity are then constants. More recent investigations (De Josselin de Jong and Geuze, 1957) have shown that the viscosity of these saturated clay systems in a state of flow is of an even more unique nature than assumed previously. Arbitrarily chosen programs of loading and unloading by varying the deviator of stress accordingly resulted in proportional rates of flow. This evid­ ence indicates that flow is a unique property of the system and that the viscosity is a constant parameter over a certain range of the time-scale. Later investigations (Geuze, 1960) have showed proof of the fact, that these systems also possess elastic properties in the lowest range of stress. Complete reversibility was obtain­ ed at programs of loading and unloading with increasing magnitudes of the deviator of stress. This behaviour proves the absence of permanent parts of the strain, which are an indication of the slip of particles with respect to each other (Tan Tjong Kie, 1954). It therefore strongly supports the hypothesis of bonds existing between the particles, which can be overcome only at a certain level of the shear stress. Two other observations are significant in this respect. Identical loading programs repeatedly performed on the same clay sample did not show essential differences in the stressstrain behaviour. The maximum level of the elastic stress deviator did not vary substantially at repeated performances provided that the permanent strain at that level was kept under control. The transition from the completely reversible state of strain to those of the partly reversible state clearly showed. The above mentioned facts have proved beyond doubt that the clay system has elastic properties within a certain limit of shear stress. The reasons why these properties did as yet not clearly show up in the conventional types of shear test has to be ascribed to the low level of the yield stress and the lack of information on the reversibility of the strains below that stress level. The sharp transition in the strain behaviour is undoubtedly due to the application of the stress deviator as a loading system, which mobilizes the bonds oriented in the principal direction of shear simultaneously. As pointed out at an earlier opportunity (Geuze, 1948) the conventional type of loading in triaxial testing involves a rotation of the principal direction of shear, which in turn mobilizes the frictional forces bet­ ween varying portions of the particles of a granular system according to the directions of their contact planes. It is an interesting point, that the yield limit of the elastic range will give us an opportunity to give the often misused term cohesion its proper physical significance, instead of its current geometrical meaning as the (assumed) point of inter­ ception of the Mohr envelope with the shear stress coordin­ ate axis. An additional point about the strain-time behaviour at instantaneous loading is the retardation effect by the viscos­ ity of the pore fluid as shown at the higher stress levels within the elastic range. This phenomenon repeats itself at any state of deformation involving the change of position of the

particles. We can as yet not make a clear distinction between retardation effects due to the fluid viscosity and the structural viscosity of the particle system. It is obvious — as stated in the early part of this discussion — that their mutual inter­ ference in the state of flow, has to be investigated in order to establish the effect of the fluid phase on the viscous proper­ ties of the soil structure. We can not however hope to obtain this information by simply removing the water from the voids or through its displacement by another type of fluid with a larger or smaller viscosity, as the magnitude of the bond forces and their mechanical nature largely depend on the nature of the fluid. The limited amount of information obtained from a cer­ tain number of clays puts the yield limit at about 50-90 gr./cm2. Beyond this limit permanent strains will appear. They are an indication of the permanent changes of the soil particle struc­ ture. The increase of the permanent deformation with time at a constant magnitude of the stress represents the state of flow, which depends on the viscosity of the soil structure. No information is as yet available on the effect of the pore pressure on the rate of the deformation. As shown by Casagrande and Wilson (1954) flow may be followed by failure at a constant magnitude of the stress. In summarizing their results these authors concluded, that the lower the stress the longer the period of time to failure. In one particular case the failure stress could be lowered by as much as 80 per cent in comparison with the failure strength at the normal rate of deformation, when a sufficiently long time to failure was allowed for. This statement is significant because it indicates that below a certain rate of loading the failure condition is much stronger affected by the permanent part of the strain than by the stress level. The ambiguity of this behaviour may well be ascribed to the interference of what is commonly known as thixotropy (Mitchell 1960), which lowers the resistance of the soil struc­ ture at high rates of strain and thereby shortens the time to failure period. As has already been mentioned by the General Reporter, the stress-strain-time relationships for saturated clays under conditions of no drainage have been studied in some detail. The amount of information under various conditions of drainage and for anisotropic materials is very limited indeed. By the design of the tests these results do not warrant conclus­ ions in view of the time dependent volume changes of the soil structure and its flow properties. These factors should have a substantial effect on the dissipation of the pore pressure. The fact that the simplest case of deformation, the one­ dimensional consolidation process, is obviously subjected to flow is a sufficient proof of this conclusion. The so-called secondary time-effects, which occur under conditions of residual pore-pressure gradients at ever de­ creasing rates of strain, deserve our special attention. As I recently observed, these effects may even take place in satur­ ated clays when subjected to hydrostatic states of stress, which should not produce shear in the bulk of the material. Evidently the shearing type of deformation then occurs in the microstructure of the clay. This type of test may serve as a means to study the viscosity of the soil structure by the increase of the particle bonds with time at extremely small rates of strain. These time dependent effects in the strain history of cer­ tain clays may well invalidate the relationships between the shear strength and the water content of these clays, which are mostly based on test results of the routine type where these bonds are lost at the state of failure. They would however show up in the type of test, as previously described, which allows for states of equilibrium at small strains and small rates of strain. It is this point, which I understand to be raised by 101

Dr Denissov in connection with pore pressure, occurring in systems where these bonds had developed either through the process of consolidation in nature or by a rest period after dissipation of pore pressure in the consolidation test. My own measurements of pore pressure at the base of the consolidation samples showed, that the magnitude of the pore pressure at repeated loading decreased when the rest periods after previous loading increased. The preceding statements and conclusions indicate that we have to study closely the properties of the bonds existing between the particles of a clay system, because they largely define the properties of the bulk of the material in either of the well known mechanisms used in soil mechanics to predict soil behaviour at varying states of stress. From what I gathered Dr Bishop said with reference to statements made by Dr Skempton, the effective area of contact between the particles would be significant in the explan­ ation of mechanisms of saturated clays. I do not think this to be a factor of such importance, because the bonds define the strength of clay structure even when the contact areas between the particles are negligible. I think that the magnitude of the bond forces and the nature of these bonds are of a decisive importance. His reference to properties of other building materials such as concrete f.i., does not apply as the bonds then are of a completely different nature. Also, the analysis of consolidation tests in clays should be based on the actual states of stress and on the stress-straintime characteristics of the material. When looking at this subject from the point of view of its practical importance in foundation engineering, it is obvious to me that our primary interest is to distinguish the various types of cohesive materials as to their tendency to flow over long periods of time. Some clays are particularly susceptible to the effects of flow on their strength and we should therefore be careful in handling these in the designs of foundation structures. Le Président : Before we continue with our discussion of the visco-elastic properties of soil Dr Denissov would like to say a few words. M. D e n is s o v Dr Bishop’s speech was a very interesting one, and I have listened to it with great pleasure, but I do not agree with some of his opinions. He said that one finds pore pressure in cemented materials such as concrete, rock, etc. It is true and I know the papers of the London Conference about Pore Pressure. But it is not proved that strength of cemented materials and rocks is decreasing due to the pore pressure increases. In my speech I have spoken about the overconsolidated clays in natural slopes. In this clay we found pore pressure, but I do not think that the fluctuations of porepressure in natural conditions can influence the stability of slopes. Therefore the pore-pressure increase cannot be the reason for essential decreasing of the safety number of clay slopes in natural conditions. I have very little time so I shall say only that it is quite possible to understand many soil phenomena without taking into consideration the influence of pore pressure. The increase of clay strength with accompanies the pressure increase is not due to the pore pressure dissipation. The reason for this is increase of cohesion due to the particles’ concentration in the unity of volume. It is very important that this considerable gain in strength remain after the un­ loading of the clay. 102

Le Président : Dr Bishop wishes to speak on the subject of the visco­ elastic properties of soil, but 1 think only to make one comment. M. B ishop Perhaps I could add one comment from a practical point of view on the dissipation of pore pressure in full-scale prob­ lems. The dam to which I referred earlier in this discussion has been built and is behaving satisfactorily. One of the reasons for our use of dissipation of pressure as a construction tech­ nique was that an earlier dam of very similar clay in an area of similar rainfall had slipped during construction when a little over half the height proposed for the Selset Dam (Banks, 1948). No special provision for the dissipation of pore pressure had been made in this case. Now, whether or not the theory is right, it is evident that the dissipation of pore pressure has a very marked effect on the shear strength of clay in full scale examples such as these. Until Prof. Denissov produces contrary evidence I think we must accept that there is some validity at least in the idea that changing the pore pressure alters the shear strength and deformation properties of actual soils. I would like also to comment on what Prof. Geuze has said. He referred, as did the General Reporter, to the small amount of data we have so far on the Theological properties of soil measured in terms of effective stress parameters. One of the reasons for emphasising this point is indicated by my earlier remarks about the non-uniformity of pore pressure within soil specimens tested under undrained conditions. Tests carried out at a constant average water content ■ — these have figured very largely in the literature of rheology —■ may be merely reflecting the effects of redistribution of water within the sample, and not the basic properties of the material. We should therefore welcome the limited amount of data presented to the Conference — there is one paper from Japan, I believe — in which the long-term properties have been presented in terms of effective stress. Such data will enable us to make some estimate of the basic Theological properties of the soil. Results published by Bjerrum, Simons and Torblaa in 1958 showed that in long-term undrained triaxial tests the pore water pressure increased very considerably with the duration of the test. This many be partly a function of the true properties of the soil or it may be partly a consequence of the test procedure, but it certainly had a very marked influence on the reduction in strength of the samples. In this connection it is worth recalling that the figure Prof. Geuze gave for the reduction in strength was from tests by Casagrande and Wilson, in which about 80 per cent of the strength was lost, on a long-term basis, in one or two cases. These were, as I recall, undrained tests, and I do not think we shall find any such marked decrease in drained tests. There will be a decrease, but it will be much smaller in magnitude, otherwise most of the world would be as flat as Holland. While on this subject I would like to draw attention to a very important paper given by Dr Hvorslev to the Research Conference on the Shear Strength of Cohesive Soils organised by the American Society of Civil Engineers last year. In this paper he considered in detail the physical components of the shear strength of saturated clays — cohesion, true angle of internal friction and that part of the strength which depends on the rate of volume change at failure — and the extent to which they are time-dependent. Anyone who wishes to take a broad view of rheological properties should include this in his reading, as well as the more widespread — and sometimes misleading — literature based on the results of undrained tests.

Références [1] B , J.A. (1948). " Construction of Muirhead Reservoir, Scotland Proc. 2nd Int. Conf. Soi! Mech., vol. 2, pp. 24-31. [2] B je r r u m , L., S im o n s , N. and T o r b l a a , I. (1958). " T h e effect of time on the shear strength of a soft marine clay Proc. Brussels Conference on Earth Pressure Problems., vol., pp.148-158. [3] H v o r s l e v , M. J. (1960). " Physical Components of the Shear Strength of Saturated Clays ”. Proc. Research Conference on the Shear Strength o f Cohesive Soils, ASCE, pp. 169-273. anks

Le Président : Would you like to comment, Dr. Geuze? M . G euze

In the first place, Mr Chairman, the flatness of Holland has nothing to do with the objections of Dr Bishop. But that is not the essence of what I have in mind. Dr Bishop particularly mentioned a type of test in which drainage was allowed to take place during a long-term creep process. It is obvious that my statements were limited to the case of no drainage because it is only for that particular state that the strain-time relationship are not or hardly affected by the spherical component of the state of stress. As previously shown (Geuze, 1948) both the consolidation test and the (direct) shear test are essentially of the mixed type as in both tests the strains result from the combined action of compression and shear. If drainage is allowed to take place, we have no means to distinguish between the creep in consolidation by the spherical component of the state of stress and the creep or flow caused by its deviatoric component. This distinction was made in a hypothetical manner on the basis of results obtained with standard type equipment in my paper to the 1948 I.C.O.S.E.F. It seems however that simular results of a more dependable nature will appear in tests as previously described. Dr Bishop’s reference to the observed creep behaviour of the material under drainage conditions therefore is not valid under conditions of constant water content; I am afraid that I can not agree with his statement with regard to the tests as carried out by Casagrande and Wilson. Le Président : To conclude this opening panel discussion I think Prof. Meyerhof would like to say a few words. Le Rapporteur Général : We have listened to an interesting panel discussion in which the three main subjects suggested at the end of the General Report have been considered. Regarding the scatter of soil mechanics tests M. Marchand has mentioned that the variation in the field can be greater than in the laboratory and he drew particular attention to the difference between problems of local stability and those of general stability, and the use of soils in their natural state as distinct from soils in the compacted state. I think that some of these factors are already taken care of in engineering practice by the different margin of safety which we use in connection with these soil properties. Thus, the factor of safety in the design of earth dams is commonly of the order of 1-5, this being a problem where we use soils in the compacted state. In foundation engineering, on the other hand, we use a factor of safety of 2 to 3, since we are here concerned with soils in their natural

state in which the variability is greater and where we have to provide a similar margin of safety as in the structure carried by these foundations. The second problem dealt with the pore pressure dissipation. Here we had the interesting results of plane strain compression tests compared with those of triaxial tests. After reviewing the papers to the Conference I made the comment that we need more data on the important case of plane strain, which holds in earth dams, the stability of slopes and retaining walls and in strip foundations. It was therefore very gratify­ ing to hear Dr Bishop giving the results of his important tests. He showed that the strength under plane strain com­ pression is generally greater and the failure strain is generally smaller than in the axial symmetrical case of compression and the pore pressure is greater in plane strain than in triaxial compression. These factors have important consequences for our understanding of the fundamental properties of soils. He also drew attention to the effect of these results on foundation problems. In my paper on piled foundations to this Conference (3B/16) I mentioned the lack of such data and indicated that a difference of only about 10 per cent in the angle of internal friction is necessary for us to be satisfied with the present theories of strip foundations on sands. In other words, it is now not necessary to revise the theories, but we did not have the right soil strength data to go with these theories. As I mentioned in Germany last week, if we use the right theory with the right soil tests we get the right answer ! Under the third subject heading for discussion, Dr Geuze referred to the importance of the visco-elastic properties of soils, the effect of stress reversals on the Bingham limit and the viscosity of soil-water systems. It was interesting to hear that the Bingham limit of clays of 30 to 90 gm./sq. cm. which he mentioned, is of the same order of magnitude as the shear strength at the liquid limit. Whether this has any connection I do not know, but it was interesting to observe this simple approximate relationship. Le Président : Well, ladies and gentlemen, that concludes the panel’s contribution to this discussion. I now propose that we adjourn until a quarter past 11. Will those members of the Conference who wish to take part in the discussion later be so kind, during this interval, as to write their names and nationalities, and the subject on which they want to speak, on a piece of paper and send it up here, so that Mr Meyerhof and myself can deal with them before we resume our discussion? (La séance fut suspendue de 11 h. à 11 h. 15) Le Président : We now open the second part of our discussion. We begin with subject (a) and I will ask Mr Ranganatham to make the first contribution. M. B. V. R (Inde) The problem of scatter of soil mechanics tests can best be studied with remoulded specimens. In case of undisturbed soil samples it will be difficult, if not impossible, to distin­ guish the scatter of test results from the variability of the particular soil under test. A special technique has been adopted to prepare remoulded clay samples at the Soil Mechanics Laboratory of the Indian Institute of Science, Bangalore. The technique, very briefly, is as follows: The clay water suspension is stirred well with a half horse­ power stirrer for sufficient time to ensure thorough and uni103 anganatham

SERIES NO. , PROPERTY MEASURED j

A

VARIATION '/.

1

2

3

4

B VAR. A X

B VAR. A X

B VAR. A /

B VAR. AVER. MAX. *

FINAL THICKNESS (m s; •475 •4 70 0-33 •498 •492 0-61 •383 •373 1-32 •475 •464 1-17 0*86

1-32

FINAL M.C. ( / ) 29-64 28-78 1-47 29-57 28 76 1-39 51-44 51-27 0-17 43-21 42-01 1-4 1 l-l 1

1*47

t 2 “2 1 ton/ft? 3-48 3-54 0-85 9-30 8-65 3-62 106-5 102-8 1-77 1-45 1-35 3-57 2-45 3-62 7X^ «•*=> 2 ton/ftf 4-17 3-99 2-16 8-45 7-74 4-39 112-0 103-4 3-99 1-62 1-55 2-21 3-19 4-39 «o* ±L Oc z ~>§ 4 ton/ft2 4-75 4-70 0*54 7-92 7*34 3-80 85-86 86-5C0-37 1-65 1-5 3 3-7 7 2-12 3-80 1 _

Cc

•465 •495 3-12 •335 •3 4 5 1-47 1-092 1-162 3-11 •610 •595 1-2 4 2-24

3-lt

My

•0475 0525 4-90 •0372 •040* 4-12 0668 0733 4-64 0548 0577 2-58 4-06

4-90

IN

IN/M IN.XIO .

RECOVERY RATIO 10-23 10-20 0-15 9-43 9-72 1-51 2-37 2-46 1-86 24-OC 23-21 1-67 1-30 k soil, full efficiency 2. k filter drain > k soil, partial efficiency 3. k filter drain sg; k soil, zero efficiency Fig. 42 shows the consolidation data obtained in triaxial tests on saturated samples of sandy silty clay (decomposed volcanic rock) from the Far East. The two samples were taken close to each other in the same borehole and classif­ ication tests showed that they were almost identical in moisture content, Atterberg limits and grading. Side filter drains were placed on the three test specimens from Sample No. 492/20, but they were not used on Sample No. 492/21. All the speci­ mens were drained from one end only. Comparing the two sets of results, the filter drains on Sample No. 492/20 appear to have been only partially effective. If the specimens of Sample No. 492/21 (without filter drains) had been drained from both ends during consolidation, the parameters t100 would have been less than those in Sample No. 492/20 employing filter drains. Now, although a still faster consolidation could be achieved by using filter drains and double drainage, the filter drains clearly have poor efficiency on this soil i.e. the tests are approaching condition 3 in which the soil and filter paper permeabilities are similar. For the particular filter paper used, Whatmans No. 5, the limiting soil permeability at which filter drains usefully assist drainage is approximately 10-6 cm/sec. This is a lower perm­ eability than most engineers imagine and the tendency in the past has been to specify filter drains on almost all soils finer than clean sands. However, it is advantageous to omit the side drains whenever possible for the following reasons : O') They affect the strength of the soil, especially at low consolidation pressures. (ii) Using filter drains, it is virtually impossible to prevent trapping air between the membrane and specimen; this can affect the volume change readings during consolidation, or can increase the back pressure re­ quired to saturate the specimen because the entrapped air has to be dissolved into solution in addition to the pore air voids. (i/7) When the efficiency of radial drainage is unknown, the consolidation data cannot be used to calculate cv and k of the soil. Checking through the records of recent triaxial tests carried out in the laboratory of Soil Mechanics Ltd, there is at least one other set of data on 6 specimens of the same soil which confirms that the filter drains are ceasing to be useful in assisting drainage when the soil permeability approaches 10-6 cm/sec. Almost all soils with this permeability have been tested with filter drains. However, from Fig. 42, it may be 121

S am ple

w

0

C (

S -

00

5

, lb./ s«* in .

33-2

3 2 0

2 O

GO

3-25

3 15

O

@

3 2 0

3 1- 7

34-4

5

2 0

- ------2

*]

3

TIm«, \n

4

J

T im e

INDEX Sam ple P AR TIC LE

SIZE

D IS TR IBU TIO N

-

7

6

In

8

4 3 2 / 2 0 1

L L

P L

3 3

34

24

3 3

3 7

27

Specimen Co.«ft.

of

No .

to

.

m / C

C O N S O L ID A T IO N Sample

9

minutes

PRO PERTIES No.

4 9 2 /2

5

P R O P E R T IE S

492/21

©

No

© 032

O

I - 6 « I 0 4 1 2 «IO *

Coeift.

of

p e r m e a b i l i t y , Vcv , c m ./s e c .

Coefft.

of

c o n s o l i d a t i o n , C V| s ^ . ^ f / j e a r

076

0

c o m p r e s s i b i l i t y . m , s < ^ .fh /to n

8 0 0

14 O O

© 0 03 7 1-4 « l O 6 14 0 0

Fig. 42 Use of side filter drains in the triaxial test.

deduced that the borderline is reached when 3" x 11" size specimens with filter drains have consolidation parameters t100 in the range of 9 - 15 minutes. Soil samples in this group — 9 in all — have been examined and compared with other soils on which filter drains would, or would not, serve a useful purpose in assisting drainage. From this examination, soils in the category of borderline cases appear to have the following characteristics in common : Visually they can be described as clayey silts or perhaps very silty clays with a clay fraction in the range 10-20 per cent. Well-graded soils in this category have a slightly lower clay fraction with just sufficient clay present to act as a binder to the coarser material. Now, the relationship between the parameter /100 and the time to failure in a drained test, tf , can be calculated from the equations in Bishop and Henkel (1957) as follows : Side drains and double drainage : tf 20 Single or double drainage : tf — 8^ /100 Therefore, soil with a permeability of 10-6 cm/sec., such as Sample No. 492/21, will have a theoretical time to failure in double drainage of approximately 40 minutes. A Labor­ atory Assistant requires a testing time of at least 2-3 hours in order to calculate the results as the test proceeds. Hence, there is no advantage in using filter drains on this soil, even if a more permeable filter paper is available. In summary, there are two main points. The first point is that the side filter drains cease to be efficient in assisting 122

drainage when the soil permeability is approximately 10-6 cm/s. The second point is that the theoretical time to failure in drained tests, published in Bishop and Henkel (1957), suggests that soil at this permeability does not, for practical purposes, need filter drains, especially on the smaller sizes of test specimen. Therefore, filter drains can be safety omitted on soils with a permeability of up to 10-6 cm/sec, and this group includes soils with an appreciable clay fraction. M. C.B. C raw ford (Canada) One of the most difficult problems in soil engineering is to relate, with sufficient accuracy, the shear strength of a soil measured in the laboratory to its actual value in the field. A great step forward was made at the time of the Third International Conference when the practical usefulness of effective stress concepts of shearing resistance was widely displayed. At the Fourth International Conference these gains were consolidated by further documentation although there was some concern about the reliability of laboratory tests for certain soils (Hvorslev 1957). One of the major contributions of this, the Fifth International Conference, is a series of papers drawing attention to the uncertainties associated with deformation and volume change during shear testing. It had been common practice to neglect the effects of strain and volumes changes in computing c' and 0 ’. This did not appear to lead to serious error with remoulded soils on which

much of the fundamental shear research had been done. But with sensitive soils it soon became apparent that the devel­ opment of pore pressure during shear was associated with collapse of the clay structure and was therefore dependent on deformation of the test specimen. It followed that if deform­ ation in the test was not compatible with deformation in situ a completely misleading interpretation might be made. Tests on the sensitive Leda clay of Eastern Canada revealed that the computed 0 ' could vary from 22° to 35° depending on the interpretation of failure. It was further reasoned that the more disturbed the test specimen, the higher would be the computed value for 0 ' (Crawford 1960). In a recent series of tests on the Leda clay (Crawford 1961) it was found useful to follow graphically the variation of effective stresses on the potential failure plane. By com­ puting 0 ' at very low strains, and correcting the computed value according to the proportion of maximum shear stress applied, it was found to be relatively independent of consol­ idation pressure and of rate of strain. This computed value of 0 \ averaging about 17°, was considered to be equivalent to the Hvorslev “ true angle of friction. ” The ability of the specimen to resist shear at greater strains, even though the effective stresses decreased to less than half their original magnitude, was attributed to “ true cohesion ”. Much more work will have to be done to substantiate the concepts mentioned above but enough has been done to show that, for sensitive clays, the shear strength parameters in terms of effective stresses cannot be obtained in the usual manner. Since c' and 0 ' are not fundamental parameters, a routine effective stress analysis may depend for its accuracy on a complete compatibility of stress path, deformation, and rate of strain between laboratory test and field condition. This points up the need for a more fundamental under­ standing of cohesion and friction in natural soils. Research in mineralogy and soil physics will be invaluable but it is probable that the engineering answer will be obtained by physical testing. Chances appear good that the shearing resistance may be satisfactorily divided into two parts, one part dependent on the effective stress on the shear plane and one part independent of the effective stress. For sensitive clays, however, these components will have to be more fun­ damental than our present concepts of c' and 0 '.

rable se produit pendant la construction, c’est-à-dire que l’essai non-drainé sur un échantillon du sol naturel donne dans ce cas des résultats trop désavantageux. D ’autre part, l’utili­ sation dans le calcul de la stabilité à long terme des valeurs de la « cohésion apparente » c’ et de « l’angle de frottement apparent » = G'f\ s/(r, s — G log cs), (Tan, 1/63). For over-consolidated clays (Fig. 79) the sand-silt grains are mutually cemented by amorphous silica, sesqui-oxydes, clay-particles in edge-flat surface and flat-flat-surface con­ tacts. Here deformation can only readily be detected, as soon as the local stresses exceeding the lower yield-values disrupt the cementing agents and surpass the mutual friction bet­ ween the sand-silt particles. This concept of the yield-value /j is described by the friction element f y parallelly coupled with the models in Fig. 78.

Fig. 79 Schematic structure and rheological models for overconsolidated clays B. For Lanchow loess the rheological properties are mainly governed by the cementing bridges holding the sand silt par­ ticles in a highly porous network (porosity 1-05, water con­ tent 8-13 per cent); the main constituents of the cementing agents; soluble salts, illites, montmorillonites, amorphous silica govern the peculiar behaviour of loess towards water. In a certain range of water contents a linear stress-strain relationship has been measured (Fig. 77 b) described by the model in Fig. 80 c.

m

(i) Fig. 80 Schematic structure and models for loess. (